The Emerging Clinical Significance of the Red Cell Distribution Width as a Biomarker in Chronic Obstructive Pulmonary Disease: A Systematic Review

There is an intense focus on the identification of novel biomarkers of chronic obstructive pulmonary disease (COPD) to enhance clinical decisions in patients with stable disease and acute exacerbations (AECOPD). Though several local (airway) and circulatory inflammatory biomarkers have been proposed, emerging evidence also suggests a potential role for routine haematological parameters, e.g., the red cell distribution width (RDW). We conducted a systematic literature search in PubMed, Web of Science, and Scopus, from inception to April 2022, for articles investigating the diagnostic and prognostic role of the RDW in stable COPD and AECOPD. The risk of bias was assessed using the Joanna Briggs Institute Critical Appraisal Checklist. Significant associations between the RDW and the presence and severity of disease, outcomes (mortality, hospital readmission), and other relevant clinical parameters (right heart failure, pulmonary arterial hypertension) were reported in 13 out of 16 studies in stable COPD (low risk of bias in 11 studies), and 17 out of 21 studies of AECOPD (low risk of bias in 11 studies). Pending further research, our systematic review suggests that the RDW might be useful, singly or in combination with other parameters, for the diagnosis and risk stratification of patients with stable COPD and AECOPD (PROSPERO registration number: CRD42022348304).


Introduction
Despite the ever-increasing global public health and financial burden of chronic obstructive pulmonary disease (COPD), considerable challenges remain in the capacity of healthcare professionals to accurately diagnose the disease, particularly during the early stages, and to predict trajectories of lung function decline, acute exacerbations (AECOPD), and other adverse clinical outcomes [1]. Contrary to other pathologies, e.g., cancer, these uncertainties have curtailed the development and implementation of a personalised medicine approach to optimise disease control and improve quality of life in patients with COPD [2]. However, they have stimulated the search for novel biomarkers that may complement the clinical information currently gathered from physical examination, lung function tests, and imaging procedures [3]. The established role of infection, inflammation, and immunity in the pathophysiology and the clinical progression of COPD has led to the identification of a number of biomarkers within these pathways, e.g., eosinophils, procalcitonin, neutrophil elastase, and serum amyloid A, which can be characterised from different biological matrices, blood, saliva, or sputum [3][4][5][6]. Though some of these biomarkers, e.g., eosinophils and procalcitonin, appear particularly promising to enhance clinical and therapeutic decisions 2 of 25 in patients with stable COPD and AECOPD [7], the availability of additional biomarkers from routine laboratory tests might also further improve patient care in those settings that have limited access to expensive and complex analytical facilities. In this context, studies have reported the diagnostic and prognostic potential of other blood cell types and derived parameters from routine laboratory tests, e.g., the neutrophil-to-lymphocyte ratio and the platelet-to-lymphocyte ratio, in patients with stable COPD and AECOPD [8][9][10][11]. A further routine haematological parameter currently under investigation as a potential biomarker is the red cell distribution width (RDW), an index used in the assessment of red blood cell morphology and anaemias that is calculated by dividing the standard deviation of red blood cell volumes by the mean corpuscular volume of the erythrocytes, and then multiplied for 100 [12,13]. Less commonly, another measure, the standard deviation of the RDW (RDW-SD), is used to express the arithmetic width of the red blood cell distribution curve measured at the 20% frequency level [14]. Elevations in RDW values, expressed as a percentage, indicate an increased heterogeneity in the volume of circulating red blood cell, a condition known as anisocytosis. The latter can be caused either by a reduced clearance of older red blood cells and/or by an impaired red blood cell production in the bone marrow [12,13]. Though an increase in the RDW has been reported in several physiological and pathological conditions [12,[15][16][17][18][19][20][21][22][23], studies conducted over the last decade have also investigated the diagnostic and the prognostic role of the RDW specifically in patients with COPD. The aims of this systematic review were: (a) to critically appraise the available evidence regarding possible differences in RDW values between patients with stable COPD and AECOPD and control groups; and (b) to investigate the associations between the RDW and clinical outcomes and other relevant clinical parameters in these groups.

Materials and Methods
We conducted a systematic literature search for articles published in PubMed, Web of Science, and Scopus, between inception and the 15 April 2022, using the following terms (and their combination): "RDW" or "red blood cell distribution width" and "COPD" or "chronic obstructive pulmonary disease" or "AECOPD" or "acute exacerbation chronic obstructive pulmonary disease". Abstracts and, if relevant, full articles were independently reviewed by two investigators. Handsearching citation lists of relevant articles was also performed to identify additional studies. The inclusion criteria for study selection were: (a) full-text available; (b) English language used; (c) description of associations between the RDW and either diagnosis, clinical outcomes, or other relevant clinical parameters in patients with COPD and/or AECOPD. The following parameters were captured from each study: age, sex, year of publication, country where the study was conducted, study design (prospective or retrospective), criteria use for the diagnosis of COPD, main comorbidities, established parameters for the assessment of the RDW in the general population (i.e., mean corpuscular volume, haemoglobin, vitamin B 12 , and folic acid), main results, and other significant associations between the RDW and clinical parameters. The risk of bias was assessed using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for case-control studies. Studies satisfying ≥75% of checklist items were considered as having a low risk [24]. The study followed the PRISMA 2020 statement on the reporting of systematic reviews (see Supplementary Tables S1 and S2) [25]. The protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO registration number: CRD42022348304).

Results of Individual Studies and Syntheses Presence of COPD
Four studies reported the presence of significant differences in RDW values between patients with and without COPD [26,28,30,62]. Sincer et al. investigated 39 patients with stable COPD of at least ten years' duration from three different centres and 39 age-and sex-matched healthy individuals without overt cardiovascular disease or end-organ injury. RDW values were significantly higher in patients with COPD (16.1 ± 2.5% vs. 13.6 ± 1.3%, p < 0.001). In multivariate logistic regression, the RDW was also independently associated with the presence of right ventricular dysfunction on echocardiogram in patients with COPD (OR = 2.20, 95% CI 1.14 to 3.86, p = 0.017) after adjusting for re-hospitalisation, forced expiratory volume in the first second (FEV 1 ), COPD stage, and FEV 1 /forced vital capacity (FVC). A cut-off RDW value of 17.7% predicted the presence of right ventricular dysfunction with a sensitivity of 70% and a specificity of 93.1% [26]. Gunay et al. retrospectively assessed 178 patients with stable COPD and 50 age-and sex-matched healthy controls. RDW values were significantly higher in patients with COPD (14.6 ± 2.2% vs. 13.7 ± 2.2%, p < 0.05) [62]. Yasar et al. investigated 140 patients with COPD, of whom 63 also had metabolic syndrome, and 50 healthy subjects. The RDW values of COPD patients were significantly higher than controls (14.3 ± 1.4% vs. 11.9 ± 1.0%, p = 0.005). However, no significant differences in RDW values were observed between COPD patients with and without metabolic syndrome [28]. Ozgul et al. prospectively investigated 175 patients with stable COPD and 210 healthy controls. Seventy COPD patients (40%) and 23 controls (11%) had RDW values > 15.5%, and COPD patients overall had significantly higher RDW values than controls (15.0 ± 2.3% vs. 13.8 ± 2.5%, p < 0.001). In multivariate analysis, the RDW was independently associated with the presence of cardiovascular disease (p = 0.01) and right ventricular dysfunction (p = 0.02) in COPD patients. A cut-off RDW value of 16.9% yielded a sensitivity of 78% and a specificity of 88.7% for right ventricular dysfunction [30]. By contrast, three studies failed to report the presence of significant associations between RDW values and the presence of stable COPD [34,36,39].

Other Clinical Endpoints
Three studies reported significant associations between the RDW and pulmonary hypertension [30,32,41]. In 175 patients with stable COPD, Ozgul et al. reported a significant and positive association between the RDW and the presence of pulmonary hypertension on echocardiogram (r = 0.1, p = 0.02) [30]. Yang et al. investigated 174 hospitalised patients with stable COPD without pulmonary hypertension, and 39 COPD patients with pulmonary hypertension. In multivariate analysis, the RDW was independently associated with the presence of pulmonary hypertension (OR = 1.521, 95% CI 1.001 to 2.313, p < 0.05) after adjusting for brain natriuretic peptide (BNP), white blood cell count, sex, haemoglobin, and pulmonary artery-to-ascending aorta ratio. The AUC value for RDW was 0.749 ± 0.054. A RDW cut-off value of 14.65 yielded a sensitivity of 69.2% and a specificity of 82.8% for pulmonary hypertension [32]. Similarly, Wang et al. reported that, in multivariable logistic regression, the RDW was independently associated with pulmonary hypertension (OR = 1.054, 95% CI 1.008 to 1.103, p = 0.02) after adjusting for GOLD stage, emphysema, PaCO 2 , NT-pro-BNP, and neutrophil-to-lymphocyte ratio. Based on the variables entered in the logistic regression, the authors also constructed a nomogram to predict the probability of pulmonary hypertension. In the predictive model, the pooled AUC was 0.770 (95% CI 0.719 to 0.820) in the training set, and 0.741 (95% CI 0.659 to 0.823) in the validation set [41]. Bai et al. retrospectively investigated 229 COPD patients, including 69 with pulmonary heart disease. In multivariate regression, RDW values were independently associated with pulmonary heart disease (OR = 1.371, 95% CI 1.197 to 1.571, p < 0.001) after correcting for white blood cell count, platelet count, mean platelet volume, platelet distribution width, and C-reactive protein. A cut-off RDW-SD value of 48 fL yielded a sensitivity and specificity of 66.7% and 99.4%, respectively, for pulmonary heart disease (AUC = 0.845, p < 0.001). Combining the RDW-SD with mean platelet volume increased the AUC value to 0.900 (95% CI 0.846 to 0.954, p < 0.001), with a sensitivity and specificity of 76.8% and 99.4%, respectively [33]. Wang J et al. retrospectively studied 125 patients to investigate the relationship between RDW-SD and pulmonary embolism. In multivariate analysis, the RDW-SD was independently associated with pulmonary embolism (OR = 1.188, 95% CI 1.048 to 1.348, p = 0.007) after adjusting for albumin, alanine aminotransferase, aspartate aminotransferase, lactate dehydrogenase, and D-dimer. A cut-off RDW-SD value of 44.5% fL yielded a sensitivity and specificity of 80.0% and 64.7%, respectively, for pulmonary embolism (AUC = 0.737, 95% CI 0.635 to 0.839). Furthermore, the combination of RDW and D-dimer further increased the predictive ability, with an AUC value of 0.897, a sensitivity of 87.5%, and a specificity of 83.5% [35].

AECOPD Severity
Two studies investigated the association between the RDW and AECOPD severity [60,62]. Koç et al., in 160 hospitalised patients with AECOPD, reported that the RDW was significantly higher in those requiring admission to the intensive care unit compared to those managed in general wards (16.9 ± 3.3% vs. 15.4 ± 2.3%, p = 0.003) [60]. By contrast, Gunay et al. did not observe any significant differences in RDW values across stage I to IV patients with AECOPD (p = 0.913) [62].

Mortality
Seven studies investigated the relationship between RDW and mortality in AE-COPD [42,[46][47][48]53,55,60]. Rahimirad et al. investigated 225 AECOPD survivors and 75 non-survivors during hospitalisation. In multiple logistic regression, the upper RDW-SD quartile was independently associated with in-hospital mortality (OR = 1.025, p = 0.002) after adjusting for age, sex, haemoglobin, and white blood cell count. An RDW-SD cut-off value of 46 fL (upper limit of normal RDW-SD) yielded an AUC value of 0.663 (95% CI 0.597 to 0.729), with a sensitivity of 64% and a specificity of 58% for in-hospital mortality [42]. Epstein et al. conducted a population-based retrospective cohort study on 539 AECOPD patients. The RDW was higher in patients with a composite endpoint of readmission or death at 60 days (15.4 ± 1.9% vs. 14.8 ± 1.8%, p = 0.0002). The association between the RDW and the composite endpoint remained significant in multivariate analysis (OR = 1.83, 95% CI 1.22 to 2.74, p = 0.0035) after adjusting for acidosis, heart failure, and Charlson comorbidity index. The AUC value, using a cut-off of 14.3%, was 0.61 (95% CI 0.57 to 0.65), with a sensitivity of 68.9% and a specificity of 48.9%. The negative predictive value of normal RDW values was 80.1% (95% CI 74.4 to 85.1) [46]. A total of 1078 AECOPD patients were studied by Torabi et al. In multivariate regression, the RDW was independently associated with in-hospital mortality (OR = 1.02, 95% CI 1.01 to 1.03, p < 0.0001) after correcting for white blood cell count, polymorphonuclear leukocytes, mean platelet volume, FEV 1 , and pulmonary artery pressure. The AUC was 0.62, with a sensitivity of 68.9% and a specificity of 48.94% [47]. In a prospective study, Hu et al. investigated the association between RDW and one-year mortality in 442 patients with AECOPD (411 survivors and 31 non-survivors). The RDW was independently associated with mortality (HR = 1.64, 95% CI 1.08 to 2.50, p = 0.022) after adjusting for age, body mass index, %FEV 1 , coronary heart disease, heart failure, renal dysfunction, blood pH, pO 2 , and pCO 2 [48]. In the study by He et al. in 132 AECOPD patients, the RDW was independently associated with one-year mortality in multivariate analysis (OR = 1.796, 95% CI 1.212 to 2.664, p = 0.004) after adjusting for age and cancer antigen 125 (CA-125). The AUC of combining the RDW and CA-125 for mortality (0.772, 95% CI 0.690 to 0.855, sensitivity 80%, specificity 67%) was superior to that of the RDW alone (0.691, 95% CI 0.595 to 0.788, sensitivity 77.5%, and specificity 53.3%, using a cut-off value of 12.75%) [53]. By contrast, no significant associations between RDW and 30-day mortality were reported by Sato et al. [55]. Similarly, Koc et al. failed to observe independent associations between RDW and six-month mortality [60].

Other Clinical Endpoints
Five studies investigated the association between the RDW and the risk of hospital readmission and/or length of stay [46,52,57,58,60]. In 160 patients hospitalised for AE-COPD, Koc et al. did not observe any significant associations between the RDW and the risk of readmission at six months [60]. Negative results were also reported by Garcia-Pachon et al. for readmission at three months [52]. By contrast, Epstein et al. reported independent associations between the RDW and 60-day readmission due to AECOPD (OR = 2.11, 95% CI 1.17 to 3.83, p = 0.01) or non-AECOPD causes (OR = 1.72, 95% CI 1.14 to 2.6, p = 0.01) after adjusting for acidosis, heart failure, and Charlson comorbidity index [46]. Zhu et al. investigated 239 patients with AECOPD and divided them into three categories according to RDW values during the first and the fourth day after admission: normal RDW, persistently high RDW, and decreasing RDW. In multivariate analysis, a persistently high RDW was independently associated with the risk of 30-day readmission (OR = 3.45, 95% 1.39 to 8.58, p = 0.008) after adjusting for the Charlson comorbidity index, coronary heart disease, and pH. By contrast, there were no significant differences in the length of stay across the three groups [57]. The same group conducted another study in 286 hospitalised AECOPD patients divided into three groups according to RDW tertile (<12.8%, 12.9% to 13.6%, and >13.6%). In multivariate logistic regression analysis, the RDW was independently associated with prolonged hospitalisation (OR = 1.400, 95% CI 1.150 to 1.703, p = 0.001) after correction for age, white blood cell count, neutrophil count, eosinophil count, FEV 1 %, acidosis, hypercapnia, diabetes, and heart failure. A cut-off value of 13.35% yielded an AUC value of 0.818 (95% CI 0.769 to 0.868), with 83.8% sensitivity and 71.6% specificity for prolonged hospitalisation [58]. Karampitsakos et al., in 160 hospitalised patients with AECOPD, reported that the RDW was independently associated with the need for non-invasive mechanical ventilation (beta coefficient, 0.054; standard error, SE, 0.025; p = 0.03) and a history of long-term oxygen therapy (beta coefficient, 0.098; standard error, SE, 0.038; p = 0.01) after adjusting for FEV 1 , FVC, haemoglobin, white blood cell count, and eosinophil count [50]. Long et al. prospectively investigated 307 AECOPD patients (195 with and 112 without symptoms of depression and/or anxiety) and 186 healthy controls. RDW values were significantly higher in AECOPD patients with depression and/or anxiety compared to AECOPD without depression and/or anxiety and healthy controls (14.2 ± 1.8% vs. 13.7 ± 1.4% vs. 13.0 ± 1.1%, respectively, p < 0.001 for trend). The AUC for depression and/or anxiety in the AECOPD group was 0.570 (95% CI 0.513 to 0.626) using an RDW cut-off value of 14% (sensitivity, 42.0%; specificity, 72.3%) [51]. Tian et al. retrospectively studied 688 patients with AECOPD without pulmonary hypertension and 206 with pulmonary hypertension. In logistic regression, no independent associations were observed between the RDW and pulmonary hypertension (p = 0.77) [56]. Finally, Peng et al. retrospectively assessed 262 AECOPD patients with (n = 80) and without (n = 182) pulmonary embolism. In multivariate regression analysis, the RDW was not independently associated with pulmonary embolism (p = 0.152) [61].

Discussion
Our systematic review identified 37 studies investigating the potential clinical role of the RDW in patients with stable COPD and AECOPD. Overall, the available evidence suggests that this routine haematological parameter used for the diagnosis of anisocytosis may not be particularly useful to discriminate between patients with stable COPD and subjects without COPD. However, it may be a useful biomarker to predict adverse outcomes in patients with stable COPD, with four studies reporting independent associations with mortality [27,29,38,40]. Furthermore, the RDW showed the potential to discriminate between patients with and without AECOPD, with seven out of eight studies reporting significant between-group differences in RDW values [43][44][45]49,54,56,59,62], and to predict adverse outcomes in AECOPD, with five out of seven studies reporting significant independent associations with mortality [42,[46][47][48]53,55,60].
The normal volume of a red blood cell typically ranges between 80 fL and 100 fL. However, this value can increase to approximately 150 fL (macrocytosis) or decrease to less than 60 fL (microcytosis) [12,13]. The coexistence of circulating red blood cells with a wide range of volumes is assessed by measuring the RDW, and conventionally defined as anisocytosis. The presence and/or the prognosis of several disease states have been shown to be associated with various degrees of anisocytosis and relatively high RDW values, particularly anaemias secondary to nutritional deficiencies (iron, folic acid, and vitamin B 12 deficiency), atherosclerotic cardiovascular disease, heart failure, atrial fibrillation, venous thromboembolism, cancer, diabetes, renal failure, liver disease, and lung disorders [12,[15][16][17][18][19][20][21][22][23][63][64][65][66][67][68]. Notably, the absolute RDW values and differences between patients with stable COPD and AECOPD and control groups in the studies identified in our systematic review are similar to those reported in other disease states, e.g., cardiovascular disease. For example, in a study by Lippi et al., the RDW values in patients with and without acute coronary syndrome were 15.1% vs. 13.5% (p < 0.001) [69] Similarly, Ma et al. reported that patients with angiographically documented coronary artery disease had significantly higher RDW values than subjects without (13.0% vs. 12.7%, p = 0.001) [70].
The results of our systematic review expand the potential clinical applications of the RDW to COPD; however, the exact mechanisms responsible for the development of anisocytosis in this group, particularly patients at risk of adverse outcomes, remain elusive. By and large, as previously discussed, anisocytosis is secondary to a reduced clearance of older red blood cells by macrophages located in the spleen and the liver and/or by a reduced production of red blood cell in the bone marrow [12,13]. Though the role of an impaired function of splenic macrophages has been poorly investigated in the context of COPD, there is increasing evidence that the liver is the primary organ involved in red blood cell clearance [71]. No-tably, liver disease is particularly common in patients with COPD, and the consequent impairment in red blood cell clearance might theoretically account for the higher RDW in this group [72][73][74]. However, in our systematic review, liver disease represented a significant comorbidity only in one study [38]. An impaired production of red blood cells, with a consequent increase in RDW values, would be expected in the context of anaemia secondary to iron, vitamin B 12 , or folic acid deficiency [12,13]. However, reduced haemoglobin was documented in only eight studies identified in our review [38,46,48,50,[55][56][57][58], and no information regarding haemoglobin status was provided in the other eight [29,33,37,39,43,45,[51][52][53]. Furthermore, data on the mean corpuscular volume were only provided in seven studies [26,27,35,42,48,55,59], whereas information regarding vitamin B 12 and folic acid status was limited to three studies [26,30,54]. Notably, other conditions that are commonly associated with an increased RDW, i.e., cardiovascular disease and diabetes, were reported in 19 out of 37 studies [26,27,29,[31][32][33]35,36,38,40,42,46,48,[56][57][58][59][60][61].
A common denominator in patients with COPD, atherosclerotic cardiovascular disease, and diabetes is the presence of local and systemic oxidative stress and increased inflammation [75][76][77][78][79][80][81][82]. Notably, there is evidence that inflammatory cytokines can inhibit the maturation of erythroid cells in the bone marrow, with the consequent release into the circulation of immature erythrocytes. This leads to an increased heterogeneity in the volume of circulating red blood cells, and higher RDW values [83,84]. The potential link between inflammation and anisocytosis is supported by the significant and positive associations reported in four studies in our systematic review between the RDW and the inflammatory markers, C-reactive protein, and neutrophil-to-lymphocyte ratio (Tables 1 and 2) [27,30,54,62]. Furthermore, a state of increased oxidative stress in red blood cells can alter their mechanical properties and structure, consequently favouring an increase in the RDW [85,86]. Though these data suggest that an increase in the RDW in COPD might reflect a combined state of respiratory and extra-pulmonary organ dysfunction, with an associated state of increased inflammation and oxidative stress, more research is warranted to investigate the mechanisms involved in the reported increase in the RDW in patients with COPD. At the same time, larger, appropriately designed prospective studies are also needed to confirm the overall results of our systematic review, particularly in terms of predicting adverse clinical outcomes. Ideally, such studies should investigate COPD and AECOPD patients with different clinical severity, comorbid status, and ethnicity, and investigate the predictive capacity of the RDW, singly or in combination with other biomarkers or scoring tools, over relatively long follow-up periods. In this context, three studies in our meta-analysis have reported that the combination with other biomarkers significantly increased the AUC value of the RDW [33,35,53].
The strengths of our systematic review include the comprehensive assessment of the diagnostic and the prognostic capacity of the RDW, as well as the strength of the association with other relevant clinical parameters, in patients with both stable COPD and AECOPD. Furthermore, the risk of bias was considered low in the majority of studies, 22 out of 27. Limitations include the lack of assessment by meta-analysis, given the heterogeneity of the studied populations, endpoints, and their characterisation; issues with the generalisability of the results secondary to the paucity of data from African, European, and American populations; and, as previously discussed, the lack of information provided in most studies regarding the presence of anaemia, vitamin B 12 , and folic acid status, important determinants of anisocytosis per se.

Conclusions
Our systematic review has shown the potential clinical role of the RDW, a routine haematological parameter that indicates the presence of anisocytosis, in discriminating between COPD patients with and without AECOPD, and, particularly, in predicting adverse outcomes in patients with COPD and AECOPD. Appropriately designed prospective studies are warranted to confirm these findings and justify the introduction of the RDW for routine clinical use in the management of COPD.